Setting Up and Other Testing Details
Setting up the tests was going to be an interesting feat. There were many issues to overcome like:
- How do I keeping the systems away from falling snow?
- How do I get my LCD display and myself warm enough to function?
- How am I going to get an Ethernet cable long enough so I can test Half-Life 2 outside?
To shelter our setup from falling snow, and attacks from neighbourhood kids wielding snowballs, testing will be performed inside the sanctity of my garage. The temperature is a little warmer in the garage, but it is still very cold (-14C to -7C at night).
Everything nearly set up in my garage.
Using my KVM and a network router as couplers, I fashioned extra long cables that stretched long enough so I can sit inside my house without having to be outside in the cold garage. My LCD would be able to function and I can be toasty warm enough to type the keyboard.
Overclocking and Testing
Overclocking was done by first establishing the highest overclock on the Asus P5GDC-V. Using the highest overclock setting from the Asus as a starting point, we then overclock the ABIT AG8 and the Chaintech V915P and adjust voltages and frequencies up or down until we achieve something reasonably stable.
To test stability of a particular overclocking setting, we run the following benchmarks:
- SiSoft Sandra 2005 — CPU Test
- SiSoft Sandra 2005 — Memory Test
- PCMark 04
- 3DMark 05
- Half-Life 2
If any of the above tests fail, then we adjust the settings again and re-run the tests. Long-term stability was not tested due to rising temperatures in the garage during the day which may affect results.
- All boards were updated to the latest available BIOS at time of testing.
- Temperatures were taken using manufacturer’s monitoring software. In the case of Chaintech, Motherboard Monitor was used.
- Testing was performed over a span of 3 nights with roughly the same ambient temperature.
- The fan on the ABIT Radeon X700 Pro died (only after 2 months of use!), so it was replaced by a 50mm 20CFM fan.